Detection of Breast Cancer using Quasi-Cross Focused Transformer-Guided Quantum Network with Lotus Optimization

Patient outcomes improve if breast cancer is discovered early. Although many deep learning methods have been studied for breast cancer identification, the efficacy of many has been limited. It presents a new method encompassing the Quasi Cross Focused Transformer Guided Quantum Network with Lotus Optimization (QCFT-GQNet-LO) which is novel to resolve these issues. This technique uses Quasi-Cross Filtering (QCF) for image pre-processing, which improves image quality by reducing noise and artefacts. It was evaluated on histopathological images from the MIAS (Mammographic Image Analysis Society) dataset. A Focused Transformer with U-Net (FTUNet) is utilized to precisely segment malignant regions, guaranteeing precise identification of crucial places. The Self-Guided Quantum Network with Triple Attention (SGQNet-TA) controls classification and feature extraction, significantly improving the model's diagnostic abilities. The whole system is optimized using the Lotus Effect Optimization Algorithm (LEOA) to improve the efficiency and robustness. The QCFT-GQNetLO approach is shown to outperform other approaches, with experimentally demonstrated accuracy of 99.9%, and recall rate of 99.8% on the MIAS dataset. This significantly improves diagnostic precision of lesion type categorization and early breast malignancy diagnosis in histopathological images.